Tag: NVidia

Posted on

Build it like I’m starting something new

How did I get here?

First, I need to explain myself. The purpose was to find out how well a PC from when I started my YouTube Channel would do now. That would be 2018, so I gathered my parts. My actual build still existed in one form, but it had upgrades, so I needed to make adjustments.

When I began my channel in 2018, I had a Ryzen 7 1700x, an X470 Asus motherboard, 16GB of DDR4 running at 2400MTS, a 1TB HDD, and an RX480, all in a Lepa 502 case. (below). Strangely, I do have all of the components, but they are spread out or put away.

lepa case
Lepa 502

What I DID have on hand, were most of the pieces and instead of the RX480, a GTX 1070 graphics card. I did have to make some changes. I no longer use the HDD as a boot drive, but this was a test to see how well the old system would hold up to modern gaming. The games were also on SSD.

So, I put everything back together, including the 600W EVGA power supply. I was so proud, as I clicked the power button. Nothing. Oh, I got power and the fans ramped up, but it didn’t post. Then I remembered why. I knew all of the components worked, but there was an issue with the motherboard.

Big Problems?

The AM4 platform has been around for ten years now, but some of the early motherboards abandoned support for older Ryzens with a BIOS update. The eprom didn’t have enough memory for every series. I have updated the BIOS on that X470 more than once for 3000 series, and for 5000. It wouldn’t post with a 1700X without another flash.

I had two solutions. I could reflash the BIOS to an earlier version, or use the next generation CPU. It was time for the Zen 2 processor. In this case, a six core, twelve thread 3600X. I do have a 3700X, but I wanted to keep this more of a mainstream experiment and the Ryzen 5 would be fine. I swapped in the 3600X and she came right up.

Another odd thing about this PC is that it runs on Windows 10 Education Edition. This is more of an enterprise version, and for quite a while it didn’t automatically upgrade to 11. It will also most likely be support past October of this year.

So, now what?

With the system up and running I started testing. Just how well a six year old system holds up was a pleasant surprise. Games ran smooth and benchmarks were consistent those run with the GTX 1070 and a modern CPU. Yes, modern CPUs run better and more efficiently, but this build is still solid.

The case, all be it large and with a plexiglass side panel, still has great airflow and a lot of room for larger video cards should someone choose to still use it. I don’t still have the original fans, but it does have room for 120mm or 140mm and room for an All in One liquid cooler. This thing holds up.

I then started comparing to PCs that were five or six years before this. Intel had the 4000 series, which were great, but AMD had the FX series. There were huge differences between the processors from 2013 to 2019, and AMD in particular looks to have made a lightyear scale jump.

Overall, this was a good little trip back, and except for the BIOS issue, I realized most of these components are interchangeable with brand new ones from today. CPUs are still being made on the AM4 platform and applications like FSR and XESS can make older Nvidia cards perform at very playable levels. I wonder if we will be able to say the same about current releases six years from now.

I will be posting the YouTube video soon, the link to it is here. If you like, there is a blog about another older build with a Xeon, here.

Posted on

Is the RTX3050 a great option for Budget Gaming?

How did we get here?

The simple answer is yes, but this question doesn’t have a simple answer. The target I usually set for a low budget gamer is between two hundred fifty and three hundred dollars. This video card will eat over half of that budget immediately. At between one hundred sixty and one seventy nine, this card is not a budget buy. What it does offer, however is great performance for its architecture. In particular, I’m talking about the 6GB version.

You see, NVidia in their infinite wisdom, has introduced another combination of different cards wit the same name. They have done this for years, like the GT1030 which came in DDR5, then DDR4. The GTX 1060 had a 6GB version and a 3GB. The RTX 3060 which had the standard 12GB, a Low Hash Version to prohibit crypto mining, and an 8GB version. They have even done it more recently by changing from DDR6x to DDR6 in some cards.

The RTX 3050 comes in two versions, or flavors if you prefer. The standard RTX3050 comes with 8GB of video memory, 256 more Cuda cores for processing, and a 1.78 GHz clock speed. The 6GB version runs with two less GB of video memory, the before mentioned shortage in processing power, and runs at 1.47 GHz. So, why would anyone buy the step-sibling?

Why I did it

The RTX3050 6GB version offers an advantage that the base version doesn’t. It has no need for external power. It is the closest thing NVidia has to a budget video card right now, placing it with the Arc A380 and the RX6400, though both are much cheaper. The offers from Intel and AMD respectively, also run on much less power and don’t need an external source, meaning a smaller power supply. Any of these cards can get by with a 300W power supply, maybe a little smaller, which means that Optiplex you picked up for 40 bucks, doesn’t need an upgrade and adapters.

In my case, I have a number of those old Dell PCs, some having 300W PSU’s or less. Until now, I have had to try the A380 or RX6400. Each of those has their own issue. The Arc A380 needs resizable bar, a configuration allowing data to flow more efficiently. Older gen hardware doesn’t support it, so performance suffers, horribly. The RX6400 runs on a four lane by PCIe 4.0 standard. Video cards of that era, all ran on 16 lanes. For comparison, NVMe SSDs run on four lanes. It is a sever limitation.

Some testing

The RTX 3050 6GB version runs on eight lanes, twice the bandwidth of the RX6400. It shows in testing, too. The RTX3050 consistently out performed the two other cards, sometimes by as much as double. In all cases, it was at least twenty percent better. But that was with a 12700KF, a recent CPU with PCIe 4.0 and resized bar, what about older stuff? I ran that too.

Running with a 4 core, 8 thread XEON E3-1270v3 from the $300 build , the results were all over the place with the RTX 3050 still outperforming overall, but with inconsistent numbers. I lead, sometimes by a few frames and other times by over twice as much. In all cases, it’s performance was predictable and replicable.

So, what’s next for the RTX3050?

This card will make the rounds on my testing bench for several weeks, then become part of a build. I’m not sure if that will be a budget build or something I would have put the GTX 1660 in, but I’m sure something like a micro ITX build will be very much to it’s liking and the small power requirement will also help keep the build cool. It’s not the cheapest card I’ve purchased, and it certainly isn’t the most expensive, so I’m not upset by the price, but this card really does need to retail for about twenty five dollars less than it does.

The reason it may not, is that it has a use case, like the RX6400. The AMD card is a single slot low profile that fits where other cards won’t, and although the performance suffers in older boxes, it’s still better than some alternatives. The 3050 has it’s place as well, with older machines that have the room, and it does perform well for it’s specs.

So, is it a budget card? No. Is it a card for a budget build? I’m saying yes and standing by my answer. It’s definitely a solid option for a budget minded build .

If you would like to read my other blogs, you can find them here. The video for the RTX 3050 can be found here.

Posted on

RTX4070 for $600? Is it now the worst GPU ever?

How did we get here?

The RTX4070 is NVidia’s latest Graphics card and in a place where they could have reset the market and blown the doors off of both of their comptetitors, it seems they have pulled up short. By the many reivews that have come out, it appears this GPU had performance room taken from it during manufacturing. Why? It’s also priced fifty to a hundred dollars more than maybe it should.

This isn’t the first time a company has put out a GPU that doesn’t make sense. Nvidia themselves did it with the GTX1070Ti, and AMD most recently did it with the RX6500. The 1070Ti was launched in a very tight market when GPUs could still be easily overclocked and had to be kneecapped so it stayed under the more expensive 1080 and 1080Ti. The 1070Ti was $399 where the 1080 came out for the $599 price we see for this one.

AMD’s grand effort was the RX6500 that had a PCI bus configuration that was one quarter the size of many other cards. That may be fine for newer systems running on a 4th generation high speen bus, but PCI gen3 was half the speed. I mention that because many of these cards were being used in older systems because of price. A price that by many accounts was still too high around $230. Even at $200, it was a hard sell. The RX6400 was even worse, but had a low profile option. Even with a slower transfer rate, it was better than other LP options.

The RTX4070 advantage?

The new offering from NVidia does have a few things going for it. It sips power. In a time where the 4090 needs 600 watts of power, the 4070 gets by with a max of 200. That also means it runs cooler, and can be smaller. One of the largest complaints about the higher end 4000 series is the size of the cards. Larger power draw means more heat which needs a bigger cooler. The 4070 doesn’t have that challenge.

It also does have Deep Learning Super Sampling 3, which doesn’t exist on previous series cards, or AMD at all. AMD does have a similar tool in Fidelity FX Super Resolution, negating part of that advantage, though. One thing that can’t be negated is a superior video encoder. The NVenc encoder still reigns supreme, and although newer AMD cards carry the AV1 encoder, some platforms don’t support it. If the streaming platform doesn’t support a process, then that process, no matter how good it is, may be useless. And the AV1 encoder IS a great tool.

So the only real question is the price. Is $600 too much to ask for a card that barely beats the next tier from the last generation? AMD has superior cards for a similar price if you are just worried about frames per second, but they also draw more power and lack some of the featues. Still, they are great cards, and do make more sense. Well, if any of these cards make sense at current pricing.

My Conclusion?

The RTX4070 is not the worst ever. That honor still goes to the RX6500 from AMD, BUT it’s on the list. The GTX1070Ti may be neck and neck with it, but selling this card for fifty bucks less would have probably kept it from being considered at all. Something to think about is the $500 for the RTX3070 at launch, would now be about the $600 for the 4070 today due to inflation and current market conditions. Money doesn’t go as far as it once did. Still, we want to feel like we are getting our money’s worth.

We probably are, but it doesn’t feel like it.

Check out the YouTube video here.

Back to the Blog page